Structural health monitoring (SHM) plays a pivotal role in ensuring the longevity and safety of concrete structures. Traditional methods for detecting cracks in concrete, while effective to a certain degree, suffer from limitations such as labor intensity, subjectivity, and the incapable to detect subsurface or minor cracks. This paper proposes a novel approach utilizing Bayesian neural networks (BNNs) to learn the relationship between concrete conductivity changes and crack width. The proposed approach takes account of the inherent uncertainties present in material properties and environmental conditions. Making use of the conductivity of fiber‐reinforced concrete, the proposed approach leverages the changes in electrical conductivity as an indicator of crack development. The BNNs model, integrating the predictive power of neural networks with the probabilistic insights of Bayesian inference, offers a significant advancement in crack width estimation. The proposed approach not only enhances the accuracy and reliability of crack detection with probabilistic output but also provides a robust framework for continuous monitoring of concrete structures, which is one of the key objectives in SHM. The results demonstrate the potential of using conductivity changes as a precise indicator for crack detection and the feasibility and accuracy of using BNN in addressing the challenges of uncertainty in monitoring data. This study contributes to the field by offering a probabilistic method for early crack detection under the uncertainty effect, enhancing the informed decision‐making by using SHM for structural maintenance.
Zhong et al. (Thu,) studied this question.